메뉴
Review your buffered writing into per-category frequency docs (grammar / rephrasings / idioms / verbs). Usage: /debrief
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Dev: check a target (spec / plan / working diff / code) for architectural drift from the rulebook docs/ARCHITECTURE.md — report qualitative deviations citing rule ids, and triage new ones into docs/Debt.md. Usage: /drift <target>
Push enriched vocab to Anki Desktop as flashcards and pull review progress back. Usage: /anki-sync
Add words/terms you're learning, enriched (translation, alternative translations, examples, synonyms, definition) into your native/learning languages. Comma-separated for several at once. Usage: /loot <word>[, <word2>, ...]
Use whenever you read, query, or mutate shadowling's sqlite database (shadowling.db) — any SELECT/INSERT/UPDATE/DELETE, ad-hoc SQL, dropping/clearing/resetting rows, inspecting data, or checking a migration's result. All DB access goes through a skill entrypoint, a repository method (`models/*`), or `sql.py` — never raw sqlite3. Schema/data-layer changes are NOT made here — they go through an appended migration in appdb.py; this skill only documents that rule and verifies the outcome.
Explain an expression in the language you're learning that you can't read literally — get a verdict (memorize vs learnable rule) + how to read it, saved to the decode dataset. Usage: /aha <phrase> [+ your hunch]
Remove words/terms from your vocab list. Comma-separated for several at once. Usage: /drop <word>[, <word2>, ...]
| name | debrief |
| description | Review your buffered writing into per-category frequency docs (grammar / rephrasings / idioms / verbs). Usage: /debrief |
| allowed-tools | Bash(python3 */debrief.py*) |
Run the deterministic debrief driver as a SINGLE Bash call that begins with
python3 and the full path — the only shape the granted Bash(python3 …)
permission matches (so nothing before it and no chaining):
python3 "${CLAUDE_PLUGIN_ROOT}/debrief.py"
The driver does everything itself — marks drills, triages each session's
languages, runs the five analytical specialists per session, and persists each
session's findings atomically. It streams progress live (flushed) so the run
never goes dark: a marked N drill(s); reviewing N session(s) header, then one
[i/N] <session> … OK / [i/N] <session> … OK (empty) / [i/N] <session> … ERROR <categories> line per session as each completes, then a totals line.
Relay that summary to the user as-is. If any session shows ERROR, tell the user
that a re-run of /debrief retries only the failed sessions. If the driver exits
non-zero with a config notice, point the user at /shadowling:setup. Do not add
analysis or doc contents of your own — the driver is the source of truth.